OpenAI ChatGPT and the Accelerated Spread of Biased Information

Motivated by the rise of new GPT language models and their impact on society, both realized and potential, we evaluated several potential impacts of those models, in particular bias and misinformation issues. Bias in machine learning models refers to their tendencies to make certain decisions more often than expected. Humans exhibit numerous biases that impact daily decisions, and although machine learning models can limit human decision-making biases, they can also create their own kinds of biases. However, when it comes to bias, not all kinds are negative, and some can be intentionally injected into the models to support equity. In the end, whether they occur in the real world or in the machine learning world, bias issues will always be open for debate. We view that debate becoming more and more important with respect to the explosion of OpenAI, in particular ChatGPT-3, and what it means for the future.